文摘
Clinicians often use intuitive models based on clinical experience or regression models based on population studies to plan treatment of gait-related disorders. Because such models are constructed using data collected from previous patients, the predicted clinical outcome for a particular patient may not be reliable. We propose a new approach that uses computational models based on engineering mechanics to predict post-treatment outcome from pre-treatment movement data. The approach utilizes a four-phase optimization process built around a dynamic, patient-specific gait model. The first three phases calibrate the model's joint, inertial, and control parameters, respectively, where the control parameters are weights in an optimization cost function that tracks the patient's pre-treatment gait motion and loads. The last phase predicts the patient's post-treatment gait pattern by performing a tracking optimization with the calibrated model modified to simulate the selected treatment.